Frozen Chicago river

US big freeze as seen from space

29 January 2019, 06:00 UTC–31 January 16:00 UTC.

Frozen Chicago river
Frozen Chicago river

Features of the 'Big Freeze' which affected parts of North America in January 2019 could be seen in satellite imagery.

Last Updated

21 February 2023

Published on

28 January 2019

By Jose Prieto and Jochen Kerkmann (EUMETSAT), HansPeter Roesli (Switzerland), Ivan Smiljanic (SCISYS), and Sancha Lancaster (Pactum)

In January 2019 many parts of the US experienced a major freeze, with temperatures dropping to -35°C. The Chicago River and many other areas of water ways froze and at least eight people were reported to have died.

 Zoomed view of Lake Chicago, Sentinel-3A OLCI True Colour, 31 Jan 16:00 UTC
Figure 1: Zoomed view of Lake Chicago, Sentinel-3A OLCI True Colour, 31 January 16:00 UTC

The extreme cold weather was caused by a highly-amplified upper air and jet stream wave that allowed a lobe of the polar vortex to migrate southward across southern Canada and the northern-central US — leading to an outbreak of Arctic air throughout the Upper Midwest and Great Lakes on 29–30 January.

Frozen American Great Lakes

The region of the Great Lakes in North America showed prominently in the Sentinel-3A pass on 31 January, at 16:00 UTC (Figure 1 and Figure 2).

The lakes supply relatively warm humidity to build cloud streets in the wind direction. The 10.8µm channel shows there was a strong temperature contrast between the water and the air temperature, 30°C colder.

 Sentinel-3A OLCI True Colour, 31 Jan 16:00 UTC
Figure 2: Sentinel-3A OLCI True Colour, 31 January 16:00 UTC

The natural looking image from OLCI (True Colour RGB from channels 20,16 and 10, Figure 2) shows a wide reflective boundary between land and lakes, in particular Michigan, due to the cold wind freezing the waters close to the west bank of the lakes and along the coast. Snow accumulates on the frozen waters which appear as bright area in OLCI wavelengths.

Lake Erie on the east shows a blue colouration stronger than the other lakes, due to a higher reflectance value at 680nm. The area of Chicago appears less snow-covered than land around, which could be a geographical effect or the result of buildings and snow being cleared by humans (Figure 1).

The temperatures, according to the 10.8µm channel of the SLSTR instrument (Figure 3), were higher for the Chicago area, compared with the surroundings, but much colder than the lakes. The grey shades represent temperatures from 228K (white end) to 275K (black end). Almost the whole scene is under 0°C (freezing).

 Sentinel-3A SLSTR, 31 Jan 16:00 UTC
Figure 3: Sentinel-3A SLSTR, 31 January 16:00 UTC

The images bring about questions on why the cloud formation is apparently more efficient over Lake Huron than over Lake Michigan or Lake Erie. There is a liquid water channel between the western side of some lakes, and the icy surface west of the lake — which could be related to upwelling waters.

The Sentinel-2A Natural Colour image at 10 m resolution (Figure 4) also shows that the weather was cold enough to allow the sea ice to be built along the west coast of Michigan lake. The cold advected air (with predominant SE direction, confirmed by the direction of the clouds streets) started to build the cumulus clouds, only after being advected over the relatively warm open water (after passing the ice sheets). This kind of mechanism leads to well-know lake-effect snow events, where in this case the snow precipitation would be expected in the eastern banks for the Michigan lake.

 Sentinel-2A Natural Colour RGB (zoomed in), 31 Jan 16:00 UTC
Figure 4: Sentinel-2A Natural Colour RGB (zoomed in), 31 January 16:00 UTC

Comparing GOES-16 imagery

The low temperatures and snow-covered surfaces had an impact on GOES RGB imagery of this event, in particular the Night Microphysics RGB, which shows the cold, snow-covered surfaces in a dark red colour (which normally belongs to thick, high-level clouds) instead of pink colours (Figure 6).

Image comparison

Snow RGB, 29 Jan 18:00 UTC compare1

Figure 5: Comparison of GOES-16 images showing the extent of the cold and snowy conditions over the US, 29 January 06:00 and 18:00 UTC (Credit: EUMeTrain)

This contrast can be clearly seen in the comparison of the GOES-16 Night Microphysics RGB with the Snow RGB (Figure 5). Also, in cold conditions, it is clear that low cloud detection is not possible in Night Microphysics RGB, e.g. the low clouds over the lakes.

 Zoomed-in GOES-16 Night Microphysics RGB with synop observations, 30 Jan 12:00 UTC. Credit: EUMeTrain.
Figure 6: Zoomed-in GOES-16 Night Microphysics RGB with synop observations, 30 January 12:00 UTC. Credit: EUMeTrain.

The snow-covered surfaces appear as a green colour in the newly devised Cloud Type RGB (Figure 7). This RGB, which uses the new 1.3 micron channel, is particularly useful for identifying thin high level clouds (red colours). More information on this channel can be found in the case Thin Cirrus seen in the new GOES-16 1.3 micron band.

 GOES-16 Cloud Type RGB, 29 Jan 18:00 UTC.
Figure 7: GOES-16 Cloud Type RGB, 29 January 18:00 UTC.

It was so dry and cold over the Midwest that surface features show up in the low level water vapour band (WV7.3), including the Great Lakes, Finger Lakes and the Ohio River (Figure 8).

Figure 8: GOES-16 Band 10 (7.3µm) , 30 January 00:42–17:17 UTC

After a few days of record cold, the amount of ice had increased significantly, as seen in the animation of the High/Low Cloud Discriminator developed by CIRA (Figure 9)

Figure 9: GOES-16 GeoColor, 30 January

Additional content

Cold weather outbreak across the Upper Midwest and Great Lakes (CIMSS Blog)
US weather: Hell freezes over as polar vortex triggers coldest freeze in a generation (Independent)
Polar vortex brings coldest air in a generation (CNN)